THREE WIDE PLANETARY-MASS COMPANIONS TO FW TAU, ROXs 12, AND ROXs 42B

We list the targets considered in this study in Table 1. Five of our targets (ROXs 12, ROXs 42B,⁸ DoAr 22, PDS 70, and FW Tau) were identified as candidate companions during the years 2001–2005, but then were subsequently neglected. Ratzka et al. (2005) conducted a K-band speckle imaging survey for stellar binarity in the Ophiuchus star-forming region and, by analyzing their data with shift-and-add stacking, they also identified faint candidate companions to ROXs 12, ROXs 42B A+B, and DoAr 22. During our own observations, we also found an additional faint candidate companion to ROXs 42B that was located interior to the published candidate. Riaud et al. (2006) also discovered a candidate companion to PDS 70, a member of Upper Centaurus-Lupus, while conducting a K-band coronagraphic adaptive optics (AO) imaging survey of young southern stars. Finally, White & Ghez (2001) discovered a wide candidate companion to FW Tau A+B while obtaining broadband optical colors of known young binary systems in Taurus-Auriga with the Hubble Space Telescope (HST); the candidate was represented by marginal (and red) detections at R_C and I_C, plus a significant detection in the narrowband Hα filter, indicating that it had significant Hα emission.

Table 1. Host Stars for Candidate or Confirmed PMCs

Name	2MASS J	$m_{r^{\prime }}$	mJ	mH	mKs	mW1	mW3	μ	SpT	AV	Dist	Teff	Mbol	log (τ)	M	Refs
				(mag)				(mas yr−1)		(mag)	(pc)	(K)	(mag)	((Myr)	(M☉)
New Companion Hosts
FW Taua	04292971+2616532	15.31	10.34	9.68	9.39	9.19	8.79	(8.9, −28.1) ± 3.3	M4	2.1	145	3270$^{+70}_{-70}$	6.52 ± 0.26	6.26 ± 0.18	0.28 ± 0.05	1
ROXs 12	16262774–2527247	15.80	11.02	9.93	9.21	8.43	5.99	(−9.5, −30.0) ± 3.4	M0	4.5	120	3850$^{+100}_{-70}$	5.94 ± 0.26	6.88 ± 0.19	0.87 ± 0.08	2
ROXs 42Ba	16311501–2432436	13.58	9.91	9.02	8.67	8.45	8.23	(−8.8, −14.6) ± 3.0	M0	2.4	120	3850$^{+100}_{-70}$	5.74 ± 0.26	6.83 ± 0.18	0.89 ± 0.08	2
Known Companion Hosts
DH Tau	04294155+2632582	13.16	9.77	8.82	8.18	7.11	5.69	(7.1, −17.9) ± 3.4	M1	1.4	145	3705$^{+70}_{-70}$	5.21 ± 0.26	6.21 ± 0.17	0.82 ± 0.09	3
GQ Lup	15491210–3539051	11.22	8.61	7.70	7.10	6.07	4.34	(−15.1, −23.4) ± 2.7	K7	1.5	155	4060$^{+150}_{-100}$	3.67 ± 0.26	6.09 ± 0.12	1.37 ± 0.10	4
1RXS J1609	16093030–2104589	12.11	9.82	9.12	8.92	8.77	8.72	(−11.2, −21.9) ± 1.5	K7	0.9	145	4060$^{+150}_{-100}$	5.19 ± 0.27	6.79 ± 0.21	1.08 ± 0.08	5
UScoJ1610-1913a	16103196–1913062	12.84	10.03	9.26	8.99	8.70	8.53	(−9.8, −22.5) ± 1.9	K7	1.9	145	4060$^{+150}_{-100}$	5.12 ± 0.27	6.76 ± 0.20	1.09 ± 0.08	6
GSC 6214-210	16215466-2043091	11.94	10.00	9.34	9.15	9.08	8.96	(−18.6, −32.2) ± 1.7	K7	0.2	145	4060$^{+150}_{-100}$	5.57 ± 0.27	6.99 ± 0.20	1.01 ± 0.07	7
Neighbors of Background Stars
2M0415+2818	04153916+2818586	15.34	10.55	9.61	9.24	8.75	6.89	(11.6, −26.0) ± 3.6	M3.75	2.0	130	3305$^{+70}_{-70}$	6.24 ± 0.26	6.17 ± 0.15	0.34 ± 0.06	8
HD 27659b	04225462+2823540	8.28	...	...	7.25	7.33	6.78	(−21.3, −23.3) ± 7.6	A4	1.1	130	8500$^{+1000}_{-1000}$	1.52 ± 0.26	6.96 ± 0.20	2.14 ± 0.20	8
PDS 70	14081015-4123525	11.15	9.55	8.82	8.54	8.03	5.72	(−32.4, −28.5) ± 1.7	K5	0.0	140	4350$^{+120}_{-150}$	5.30 ± 0.26	7.08 ± 0.20	1.15 ± 0.09	9
GSC 06191-00019a	15590208–1844142	11.51	9.00	8.34	8.11	7.93	7.75	(−8.5, −24.8) ± 1.9	K6	1.4	145	4350$^{+120}_{-150}$	4.70 ± 0.26	6.78 ± 0.21	1.32 ± 0.10	7
UScoJ1608-1935	16082387–1935518	13.57	10.20	9.47	9.25	9.07	8.91	(−10.5, −25.7) ± 2.3	M1	1.4	145	3705$^{+70}_{-70}$	5.65 ± 0.26	6.57 ± 0.18	0.75 ± 0.08	7
ScoPMS 42b	16102174–1904067	14.23	10.68	9.91	9.62	9.31	9.09	(−12.5, −25.8) ± 3.1	M1	1.7	145	3705$^{+70}_{-70}$	6.04 ± 0.27	6.76 ± 0.19	0.72 ± 0.08	7
GSC 06793-00994	16140211–2301021	11.07	9.38	8.77	8.61	8.41	8.35	(−7.1, −20.7) ± 0.8	G4	1.3	145	5840$^{+35}_{-35}$	4.23 ± 0.27	7.72 ± 0.19	1.20 ± 0.04	7
GSC 06794-00156a	16245136–2239325	9.40	7.78	7.28	7.08	6.99	7.00	(−11.8, −22.1) ± 1.4	G6	1.0	145	5700$^{+35}_{-35}$	3.27 ± 0.27	7.24 ± 0.07	1.35 ± 0.04	7
DoAr 22	16261932–2343205	12.18	9.73	9.25	9.02	8.82	8.62	(4.5, −20.0) ± 1.3	F5	3.1	120	6440$^{+800}_{-400}$	4.23 ± 0.26	7.89 ± 0.34	1.24 ± 0.05	2
ScoPMS 214	16294869–2152118	10.92	8.68	8.00	7.76	7.61	7.52	(−9.9, −23.9) ± 1.2	K0	1.9	145	5250$^{+170}_{-170}$	3.55 ± 0.27	7.02 ± 0.19	1.40 ± 0.05	10

Notes. ^aThe primary is itself a known binary; the observed flux ratio has been used to correct the J magnitude before computing M_bol and placing the primary of the system on the HR diagram. Binary properties were adopted from Kraus et al. (2011) for FW Tau, Ratzka et al. (2005) for ROXs 42B, and Kraus et al. (2008) for UScoJ1610-1913, GSC 06191-00019, and GSC 06794-00156. ^bThe J and H magnitudes for HD 27659 are flagged as erroneous in 2MASS, so we adopt an assumed color of J − K = 0.05 (Kraus & Hillenbrand 2007b) and the K_s magnitude from 2MASS in order to calculate M_bol. References. (1) White & Ghez 2001; (2) Ratzka et al. 2005; (3) Itoh et al. 2005; (4) Neuhäuser et al. 2005; (5) Lafrenière et al. 2008; (6) Kraus & Hillenbrand 2007a; (7) Ireland et al. 2011; (8) This Work, 9) Riaud et al. 2006; (10) Metchev & Hillenbrand 2009.

Download table as:  ASCIITypeset image

Another seven of our targets (2M04153916, HD 27659, ScoPMS 42b, GSC 06191-00019, GSC 06793-00994, GSC 06794-00156, and UScoJ1608-1935) were identified to have candidate companions in the course of our own survey programs. Two of these stars (GSC06793-00994 and GSC 06794-00156) were part of the sample of Upper Scorpius (Upper Sco) members we described in Ireland et al. (2011), but their candidate companions could not be tested for association in that work. Another three stars (GSC 06191-00019, ScoPMS 42b, and UScoJ1608-1935) were observed to have faint candidate companions during the same campaign, but we did not include them in our previous analysis because they are binary systems with projected separations that impinge on the PMC semi-major axis range (ρ > 150 mas). Finally, two targets (2M04153916 and HD 27659) were identified to have faint candidate companions during a snapshot K' band AO imaging of new Taurus members recently discovered by Luhman et al. (2009) and Rebull et al. (2010); these observations were primarily intended to screen for wide binary companions in anticipation of future observations with non-redundant aperture masking.

Finally, the last six targets we consider (DH Tau, GQ Lup, 1RXSJ1609, UScoJ1610-1935, GSC 6214-210, and ScoPMS 214) were identified as confirmed or likely low-mass companions to young stars in Taurus (DH Tau; Itoh et al. 2005), Lupus (GQ Lup; Neuhäuser et al. 2005), and Upper Sco (the remaining four; Lafrenière et al. 2008; Kraus & Hillenbrand 2009b; Metchev & Hillenbrand 2009; Ireland et al. 2011). All were shown to be comoving with their neighbor star and hence are generally assumed to be associated. The only exception is the companion to ScoPMS 214, which Metchev & Hillenbrand (2009) show to have a spectral type inconsistent with the expected luminosity for a faint companion; the implication is that it is likely to be an unassociated field star with similar motion to Upper Sco. We will revisit this identification in Section 4.3.

Many of the candidate companions in our sample were initially (or also) identified by other observing programs, as were several confirmed PMCs that we are also studying. We have collated all observations for these targets in Table 2 and will combine them with our own measurements in order to test the association of candidate companions and measure a uniform set of colors for confirmed PMCs.

We have adopted most of these observations as stated in their sources, but several seem to require updated parameters. The observation of PDS 70 by Riaud et al. (2006) did not include a position angle (P.A.), so we have measured its P.A. from images downloaded from the European Southern Observatory archive. The observation of FW Tau by White & Ghez (2001) included a P.A. that is discrepant by ∼111° from our own measurements, so we also measured an updated value of the P.A. using the processed images from the Hubble Legacy Archive. In both cases, we conservatively assess an uncertainty of ∼05, corresponding to the angle subtended by the FWHM of the companion point-spread function (PSF). Finally, the observations reported by Ratzka et al. (2005) for several companions (including two comoving companions) disagree with other observations by significantly more than their reported uncertainties (which are typically <10 mas). Since they reported the detections at signal-to-noise ratios ∼5 from shift-and-add speckle imaging, then we expect that the minimum positional uncertainties should be ∼1/5λ/D or ∼30 mas. We therefore have adopted uncertainties of 30 mas for their projected separations and the corresponding subtended angle for their P.A.s.

More generally, we note that many observations seem to base their uncertainties on the scatter in their measurements, without considering systematic errors. We therefore also have adopted the systematic uncertainty floors that we assign to our own NIRC2 data (which are typically more stable than data from most other instruments). Due to uncorrected residuals from geometric distortion (e.g., Anderson & King 2003; Yelda et al. 2010), all astrometry is given a minimum uncertainty of 2 mas. Furthermore, due to the uncertain plate scale, all projected separations are given a minimum fractional uncertainty of 10⁻³. Finally, due to anisoplanatism and variability of the host stars, all contrast ratios are given a minimum uncertainty of 0.05 mag. These uncertainties could be larger for targets observed using multiple telescopes, instruments, and observing techniques. However, there has never been a global calibration of the plate scale of geometric distortion (such as with observations of globular cluster fields) for any significant fraction of high-resolution imaging instruments.

We obtained photometric data for the host stars of each candidate from publicly available all-sky surveys. The optical r' magnitudes for most host stars were taken from the 14th Carlsberg Meridianal Catalog (CMC14; Evans et al. 2002). Some targets are too far south or otherwise do not have CMC14 counterparts, so for those stars we adopted the average of the USNO B1.0 R magnitudes (Monet et al. 2003). Based on the stars that have measurements in both catalogs, the magnitudes are equivalent to within the uncertainties (R_USNOB − r' ∼ 0). The 1.0–2.5 μm (J, H, and K_s) data were obtained from 2MASS (Skrutskie et al. 2006). Most of our AO imaging measurements were taken with a K' filter (λ = 2.124 μm), while measurements from the literature often use K (λ = 2.196 μm) or K_s (λ = 2.146 μm), but the conversion terms for late-M and early-L objects are negligible (≲0.01–0.02 mag; Carpenter 2001) compared with the typical systematic uncertainties of AO photometry. Finally, the 3.0–15 μm data (W1 and W3) were obtained from WISE (Wright et al. 2010). The WISE W1 filter is centered at 3.35 μm (2.9–3.8 μm), while our ground-based measurements were taken with a Mauna Kea Observatories (MKO) L' filter centered at 3.78 μm (3.4–4.1 μm), so use of WISE data requires a color correction. The conversion has not yet been reported in the literature, so we downloaded WISE data for 40 M3.0-L9.5 field dwarfs that have L' data reported in the literature (Leggett et al. 2010) and computed a color–SpT relation of W1 − L' = 0.044 × SpT − 0.078 (where SpT = 0 for M0 and 20 for T0), which has a scatter about the relation of σ = 0.10 mag for the full sample and σ = 0.05 mag specifically for M5-L1 dwarfs. We tested this relation for the isolated BDs observed with L' photometry by Jayawardhana et al. (2003) and found that the mean W1 − L' colors agreed to within 0.2 mag, with a scatter of 0.2 mag; similar offsets were seen for disk-bearing and disk-free BDs, indicating that infrared excesses do not bias this relation.

None of the host stars were observed by Hipparcos or otherwise have direct trigonometric parallaxes, so we must infer their distances from the mean and standard deviation of other members of their associations. Based on previous trigonometric parallaxes, we have adopted these values collectively for all members of Taurus (145 ± 15 pc; Torres et al. 2009), Ophiuchus (120 ± 10 pc; Loinard et al. 2008), Upper Sco (145 ± 15 pc; de Zeeuw et al. 1999), and Upper Centaurus-Lupus (140 ± 15 pc; de Zeeuw et al. 1999). The number of Hipparcos parallaxes toward the Lupus dark clouds is small, so we instead adopt the extinction-based measurement of 155 ± 15 pc reported by Lombardi et al. (2008), which is consistent with the convergent-point distance recently reported by Galli et al. (2013). Where available, we adopt proper motions for each star that were reported in the UCAC3 catalog (Zacharias 2010). However, some targets are too optically faint for UCAC3, so we instead report proper motions derived from USNO-B1, 2MASS, and Sloan Digital Sky Survey (SDSS) using the methods described in Kraus & Hillenbrand (2007b).

We have adopted the spectral types for the host stars as reported in the literature. Many of the host stars also have extinction measurements reported in the literature. However, some are missing such measurements and others appear to be significantly redder than expected for the given spectral type and A_V. We therefore have calculated new extinction estimates by comparing each star's observed r' − J or R − J color to that for field dwarfs of the same spectral type (Kraus & Hillenbrand 2007b), converting the color excess to a visual extinction using the extinction relations of Schlegel et al. (1998). Most of our new measurements agree to within ΔA_V ≲ 1 or ΔA_J ≲ 0.25. However, two stars (ROXs 12 and DoAr 22) are significantly more extincted than reported in the literature (ΔA_V ∼ 3).

We estimated the temperatures of the host stars based on their spectral types, combined with the dwarf temperature scale of Schmidt-Kaler (1982) for ⩽M0 stars and the pre-main sequence temperature scale of Luhman et al. (2003) for >M0 stars. We estimated their luminosities from their dereddened 2MASS J magnitudes, the inferred distances listed above, and the J band bolometric corrections we previously tabulated in Kraus & Hillenbrand (2007b). Finally, we used these temperatures and luminosities to place each host star on the Hertzsprung–Russell (HR) diagram, where they can be directly compared with the predictions of pre-main sequence evolutionary models. Different models make highly discrepant predictions, but based on past calibrations (as summarized by Hillenbrand & White 2004) and the dynamical masses that we have measured in these regions (e.g., M. J. Ireland et al., in preparation; Kraus et al. 2014), we have chosen the models of Baraffe et al. (1998) using a convective mixing length of 1 H_P. The star HD 27659 falls outside the temperature range spanned by those models, so we instead used the models of Siess et al. (2000). This said, its proper motion is not consistent with Taurus membership, so an age and mass derived for the distance of Taurus might not be meaningful.

We summarize the properties of these host stars in Table 1.

Most of our high-resolution imaging observations were obtained at Keck Observatory using the Keck-II 10m telescope and NIRC2, its facility AO imager. GSC 06191-00019 was also observed at Palomar Observatory using the Palomar-Hale 200'' telescope and PHARO, its facility AO imager (Hayward et al. 2001). All of the primary stars are brighter than R = 15, so we observed them with natural guide star AO at both Keck and Palomar.

The observations span a number of observing seasons (2007–2013) and were taken by several different observers, so they vary significantly in their details (such as total integration time, dither pattern, and Fowler sampling). Several observations using NIRC2 in K' were also taken using the 600 mas coronagraphic spot, which provides ΔK' = 7.17 ± 0.10 mag of attenuation for the primary star (as measured from observations of binary stars). All observations used the smallest pixel scales (10 mas/pix with NIRC2 and 25 mas/pix with PHARO). We summarize the salient details for these observations in Table 3 and also refer readers to the Keck Observatory Archive,⁹ which now hosts all NIRC2 data after an 18-month proprietary period.

The data analysis follows the same prescription as described in Kraus et al. (2008). To briefly summarize, we measured astrometry and aperture photometry for each source with respect to the purported primary star using the IRAF task DAOPHOT (Stetson 1987); all measurements were conducted using apertures of 0.5, 1.0, and 2.0λ/D and then the optimal aperture was chosen to maximize the significance of the detection (given the competing uncertainties from the primary star's PSF halo and the Poisson noise for the candidate companion). In order to estimate the uncertainties from the data, we analyzed the measurements in individual frames and then combined those measurements to estimate the mean and standard deviation. We then corrected for geometric distortion in the images, accounting for the remaining systematic uncertainty in the plate scales and distortion solutions of PHARO (0.3%; Ireland et al. 2011) and NIRC2 (0.05%; Ghez et al. 2008; Cameron 2008) by adding those terms in quadrature with the observed scatter. Candidates were judged to be comoving if their relative proper motion fell within <3σ of zero and disagreed with the expected motion of a background star by >3σ.

In Table 3, we list the observations and results for all candidate companions that we observed.

Young stars are intrinsically variable due to several phenomena (i.e., spots, accretion, and variable extinction), so the conversion of contrasts into magnitudes and colors is subject to systematic uncertainties. Furthermore, the precision of AO photometry is fundamentally limited by anisoplanatism (e.g., Steinbring et al. 2002), especially for relatively faint targets that were observed with modest Strehl ratios. To average down these uncertainties, we have combined all available photometric measurements for each object to compute a weighted mean contrast in each filter, with an additional caveat that no epoch was given an uncertainty better than ∼0.05 mag. We then computed apparent magnitudes by adding each contrast measurement to the known magnitude for its parent star (Section 2.3; Table 1), adding the uncertainties in quadrature and also assessing a systematic uncertainty of ∼0.05 mag in the brightness of the primary (due to its potential variability at the 2MASS epoch; Carpenter et al. 2002). We report the resulting apparent magnitudes in Table 4.

Table 4. Candidate Companion Properties

Name	$M_{K^{\prime }}$	mJ	mH	$m_{K^{\prime }}$	$m_{L^{\prime }}$	μρ	μPA	ρ	M
	(mag)	(mag)	(mag)	(mag)	(mag)	(mas yr−1)	(mas yr−1)	(AU)	(MJup)
New Companions
FW Tau	9.51 ± 0.24	17.34 ± 0.07	16.25 ± 0.07	15.32 ± 0.07	14.25 ± 0.10	−1.2 ± 0.2	0.8 ± 0.7	330 ± 30	10 ± 4
ROXs 12	8.93 ± 0.19	16.64 ± 0.07	15.37 ± 0.08	14.32 ± 0.06	12.55 ± 0.09	3.4 ± 2.0	−6.5 ± 1.9	210 ± 20	16 ± 4
ROXs 42B (b)	9.62 ± 0.19	16.99 ± 0.07	15.88 ± 0.06	15.01 ± 0.05	14.15 ± 0.09	0.2 ± 1.1	−0.7 ± 1.6	140 ± 10	10 ± 4
Known Companions
DH Tau	8.34 ± 0.24	15.69 ± 0.07	14.88 ± 0.07	14.15 ± 0.07	12.93 ± 0.08	0.2 ± 0.2	−1.0 ± 0.3	340 ± 30	18 ± 4
GQ Lup	7.37 ± 0.22	14.82 ± 0.11	13.79 ± 0.08	13.33 ± 0.06	12.29 ± 0.10	−2.0 ± 0.2	2.5 ± 0.3	110 ± 10	31 ± 3
1RXS J1609	10.42 ± 0.23	17.90 ± 0.13	16.87 ± 0.09	16.23 ± 0.06	14.87 ± 0.31	−0.5 ± 0.9	−1.0 ± 1.1	320 ± 30	10 ± 2
UScoJ1610-1913	7.00 ± 0.23	13.89 ± 0.08	13.28 ± 0.07	12.80 ± 0.06	12.09 ± 0.09	1.8 ± 1.8	−1.2 ± 1.6	850 ± 90	70 ± 10
GSC 6214-210	9.15 ± 0.23	16.30 ± 0.08	15.55 ± 0.07	14.95 ± 0.06	13.83 ± 0.09	0.4 ± 0.6	−1.7 ± 0.6	320 ± 30	16 ± 1
Unassociated Objects
2M04153916	...	...	...	14.91 ± 0.07	...	...	...
HD 27659	...	14.31 ± 0.07	14.03 ± 0.07	14.09 ± 0.07	14.37 ± 0.10	−5.5 ± 4.2	7.8 ± 4.2
PDS 70	...	13.87 ± 0.07	13.17 ± 0.08	13.20 ± 0.07	13.30 ± 0.09	24.7 ± 0.3	26.0 ± 3.0
GSC 06191-00019	...	...	...	16.54 ± 0.07	...	29.0 ± 1.7	−21.2 ± 1.9
UScoJ160823.8-193551	...	...	...	16.53 ± 0.07	...	−10.4 ± 1.4	15.4 ± 2.0
ScoPMS 42b (cc1)	...	...	...	17.00 ± 0.07	...	1.3 ± 1.4	−31.9 ± 1.1
ScoPMS 42b (cc2)	...	...	...	16.39 ± 0.07	...	17.0 ± 1.8	0.9 ± 1.1
ScoPMS 42b (cc3)	...	...	...	16.79 ± 0.07	...	15.8 ± 1.8	2.9 ± 1.1
GSC 06793-00994	...	...	...	16.41 ± 0.06	...	17.4 ± 2.6	16.2 ± 3.2
GSC 06794-00156	...	...	...	16.51 ± 0.07	...	7.3 ± 3.8	14.6 ± 4.3
DoAr 22	...	15.95 ± 0.07	...	14.96 ± 0.06	...	16.0 ± 2.1	24.1 ± 2.2
ScoPMS 214	...	...	...	13.57 ± 0.06	...	−1.1 ± 0.8	−8.6 ± 0.6
ROXs 42B (cc1)	...	...	15.22 ± 0.07	15.46 ± 0.06	15.04 ± 0.12	−9.1 ± 1.1	14.4 ± 3.2

Notes. Each photometry measurement reflects the weighted mean of all measurements in Tables 2 and 3, except those that are footnoted as having been explicitly excluded. The minimum uncertainty at an epoch is assumed to be σ = 0.05 mag (due to stellar variability of the primary). The uncertainties here also reflect the uncertainty in the primary star magnitudes taken from 2MASS and WISE, as well as the color–SpT relation used to convert WISE W1 to MKO L'. Stellar variability at those epochs also imposes a systematic uncertainty of σ = 0.05 mag. The magnitudes reported here have not been corrected for extinction, but we have de-extincted or dereddened using the A_V inferred for each corresponding primary (Table 1) before computing masses or colors (i.e., Figures 7–10). The uncertainties in projected separations are driven entirely by the uncertainty in distance and hence have the same fractional uncertainty as the quoted values (Section 2.3).

Download table as:  ASCIITypeset image

We measured the relative motion of each object with a weighted linear fit to compute (μ_{rel, α}, μ_{rel, δ}). Due to the precise and accurate calibration of NIRC2 astrometry (e.g., Yelda et al. 2010), then any fit with multiple NIRC2 points (including those of the three bona fide companions) is effectively dominated by the NIRC2 measurements. If the companion was not associated, then there also would be an error term for parallactic motion, but such motions are small compared with the proper motion over timescales of >1 yr. We report the relative motions and the inferred physical projected separations in Table 4.

For the bona fide companions, the absolute magnitudes of the companions were converted into masses (in M_Jup) using the hot-start DUSTY models, which are more appropriate than the COND models due to the dusty nature of young low-temperature photospheres (Cruz et al. 2009; Allers et al. 2010; Bowler et al. 2010; Patience et al. 2010). The reported uncertainties are dominated by the primary stars' poorly constrained ages, which leave the masses uncertain by up to 50%; the photometric uncertainties are negligible in comparison. The plausibility of the models adds another significant systematic uncertainty; newer models suggest that the hot-start models could significantly underestimate planet masses (Fortney et al. 2008; Spiegel & Burrows 2012; Marleau & Cumming 2013).

In each case, we use the hot-start DUSTY models to estimate a range of plausible masses based on the age range seen in its host region (1–5 Myr for Taurus, Ophiuchus, and Lupus; 5–10 Myr for Upper Sco), and assign a best-fit mass that is the average of these minimum and maximum limits. These age ranges encompass a possible upward revision of all pre-main sequence stellar ages (e.g., Pecaut et al. 2012) and hence allow for higher masses than were previously estimated. We will discuss our estimates for the companion masses in more detail in Section 4 and list the mass ranges in Table 4.

As we summarized in Table 4, three of the candidate companions in our sample are comoving with their host stars. Each is located in a wide orbit (ρ ≳ 150 AU) and has an apparently planetary mass (M = 6–20 M_Jup). All were previously reported in the literature as candidate binary companions, but have been neglected for the past decade. Based on the two-point correlation function in unbound associations like these regions (Kraus & Hillenbrand 2008), the probability of chance alignment with an unbound association member within <3'' is small (<1 chance alignment, including stellar-mass companions, per 10⁴ members) and hence these objects represent a population of bound companions near or within the planetary-mass regime.

4.1.1. FW Tau b

FW Tau AB is located near the center of the Taurus-Auriga complex and is a close (ρ ∼ 75 mas; 11 AU) binary system comprised of two stars with near-equal fluxes at wavelengths of 0.3–2.2 μm and equal fluxes in the K' band (White & Ghez 2001). Given the unresolved system spectral type of M4 (Briceno et al. 1993), models suggest individual masses of 0.28 ± 0.05 M_☉ for each component. The corresponding isochronal age is 1.8$^{+1.0}_{-0.5}$ Myr, consistent with the median age of Taurus-Auriga (1.8 Myr; Kraus & Hillenbrand 2009a). Observations of the Hα emission line strength (EW[Hα] = −17 Å; Briceno et al. 1993) and the lack of excess emission from near-infrared (NIR) and mid-infrared photometry (Rebull et al. 2010) suggest that the primary is not accreting and does not host a substantial disk around either component or within the inner ∼50 AU. However, Andrews & Williams (2005) reported a 4σ detection of submm flux (F_ν = 4.5 ± 1.1 mJy) that could indicate a disk with a very low mass (M_disk ∼ 0.2 M_Jup) located around one or more objects in the system.

The candidate wide companion to FW Tau was first identified by White & Ghez (2001), who noted a faint object at a projected separation of ∼23 (∼330 AU) from the primary star of the central binary. This candidate companion was marginally detected by HST imaging in the F675W (R_C) and F814W (I_C) filters, as well as having a significant detection in the narrowband F656N (Hα) filter. They concluded from its apparently significant Hα line emission that it could be an accreting companion. However, no further observations of the candidate companion have since been reported.

We obtained further observations of the FW Tau system in 2008 and 2011 and, as we show in Figure 1 (left), the candidate companion has a significant (but quite faint) NIR counterpart. The multi-epoch astrometry shown in Figure 1 (right) demonstrates that the candidate is indeed a comoving companion, with a net relative motion of 1.7 ± 0.9 mas yr⁻¹ (∼1.1 km s⁻¹) with respect to the primary of the central pair. Based on its dereddened absolute magnitude of $M_{K^{\prime }} = 9.28 \pm 0.24$, we estimate a companion mass of 10 ± 4 M_Jup for system ages of 1–5 Myr (Chabrier et al. 2000). The extremely red color of the companion (dereddened J − K' = 1.73  ±  0.13) is consistent with this identification, especially given the low line-of-sight reddening to FW Tau AB (A_V = 0.4; Table 1). This very red J − K' color is typically only seen among field dwarfs for the reddest late-L dwarfs but, as was noted by Allers et al. (2010), young objects often seem to be redder than field counterparts of similar temperature. Theoretical models predict that the companion should have a temperature of T_eff ∼ 1900–2100 K and hence a spectral type of ∼L1–L2. Given its bound nature and apparently planetary mass, we hereafter denote the companion as "FW Tau (AB) b" or "FW Tau b" for simplicity.

Zoom In Zoom Out Reset image size

The significant Hα emission from FW Tau b, combined with the possible presence of a disk somewhere in the system, indicate a high likelihood that it hosts a (circumplanetary?) disk like that of GSC 6214-210 b (Bowler et al. 2011). The presence of this disk also introduces some ambiguity regarding the intrinsic luminosity of the companion, as it could be a stellar or BD companion with an edge-on disk (like HV Tau C or HK Tau B; Stapelfeldt et al. 1998; Duchêne et al. 2010). However, the morphology argues against this explanation. Stellar edge-on disks show a characteristic central dark lane that separates the light reflecting from the two surfaces of the disk and we see only a single unresolved point source.

There are two known substellar hosts of edge-on disks, IRAS 04325+2402 C (Hartmann et al. 1999; Scholz et al. 2008) and 2MASS J04381486+2611399 (Luhman et al. 2007), both with SpT ∼ M7.25. IRAS 04325+2402 C has a mass of M < 0.1M_sun, although the NIR spectrum is heavily veiled and a more precise mass determination is difficult. HST images clearly show the characteristic dark lane seen for higher-mass edge-on disks, indicating that at least some substellar edge-on disk systems follow this morphology. 2MASS J04381486+2611399 is an M7.25 BD (M ∼ 50 M_Jup) that also appears to host an edge-on disk, but is more easily characterized from its scattered light spectrum because its inner disk is cleared and hence there is no veiling or NIR excess. HST images show a bipolar outflow emerging from the BD, but there is not a clear dark lane. However, it is >2 mag brighter than FW Tau b in K'. Assuming similar amounts of attenuation in the K' flux and neglecting possible NIR excesses in FW Tau b, then the 1 Myr DUSTY models still indicate that FW Tau b would have M < 15 M_Jup if seen in a similar geometry. Nonetheless, a dispositive result would require spectroscopic detection of photospheric features that indicate the T_eff of the central object.

4.1.2. ROXs 42B b

ROXs 42B AB is a little-studied member of the Ophiuchus complex, located ∼2° east of its core. ROXs 42B was identified as a close binary system by Simon et al. (1995) and Ratzka et al. (2005); the latter survey measured a flux ratio of ΔK ∼ 1.1 and a projected separation of ρ = 83 mas (∼10 AU). As we will report in a future work, our own observations with non-redundant aperture-mask interferometry have not recovered this companion at a projected separation of ρ > 15 mas, suggesting that orbital motion has carried it inward over the past decade. The (unresolved) system was studied with optical spectroscopy by Bouvier & Appenzeller (1992), who determined a spectral type of M0 and identified it as a young star based on the presence of weak Hα emission (EW[Hα] = −1.7 Å), X-ray emission, and possible lithium absorption. Models suggest individual component masses of 0.89 ± 0.08 M_☉ and 0.36 ± 0.04 M_☉, respectively, and a corresponding isochronal age of 6.8$^{+3.4}_{-2.3}$ Myr. The system has no excess in W3 and no counterpart in W4 from WISE observations, no radio counterpart at 1.3 mm (F_ν < 45 mJy; M_disk < 9 M_Jup; Andrews & Williams 2007), and Hα emission consistent with most WTTSs and hence there is no evidence that it hosts an optically thick protoplanetary disk.

The binary survey by Ratzka et al. (2005) also identified a faint candidate companion at a much wider projected separation (∼11; ∼140 AU) in the shift-and-added stack of their speckle data. However, it has been neglected in the subsequent literature, so we observed the system in 2011 and 2012 to confirm its existence (Figure 2, left) and test for common proper motion (Figure 2, right). Our observations recovered the candidate companion, but as can be seen in Figure 2 (left), they also revealed another candidate companion even closer to ROXs 42B (ρ = 055; ∼70 AU).

Zoom In Zoom Out Reset image size

As we show in Figure 2 (right), the outer candidate is indeed a comoving companion to ROXs 42B, with a relative motion of 0.7 ± 1.6 mas yr⁻¹ (∼0.5 km s⁻¹) with respect to the photocenter of the central binary. Based on its dereddened absolute magnitude of $M_{K^{\prime }} = 9.39 \pm 0.19$, we estimate a companion mass of 10 ± 4 M_Jup for system ages of 1–5 Myr (Chabrier et al. 2000). The companion has a luminosity and colors very similar to FW Tau b, so we expect similar atmospheric and bulk properties. Given its bound nature and apparently planetary mass, we hereafter denote the companion (with some regret regarding the nomenclature) as "ROXs 42B (AB) b" or "ROXs 42B b" for simplicity.

In Figure 3, we show the corresponding proper motion diagram for the inner candidate. The nature of this object is less certain; the proximity to the bright primary leaves its colors and astrometry less reliable and the larger expected orbital motion (∼6 mas yr⁻¹ for a circular face-on orbit) is a significant fraction of the absolute proper motion of the system. Nonetheless, the measured astrometry agrees very well with the track expected for a non-moving background star, with a total relative motion of 17 ± 3 mas yr⁻¹, and its colors appear to be quite blue (H − K' = −0.21 ± 0.11, K' − L' = 0.09 ± 0.14). We therefore strongly prefer the background hypothesis over the bound companion hypothesis for this object and we provisionally designate it as a background star (hereafter, "ROXs 42B cc1").

Zoom In Zoom Out Reset image size

4.1.3. ROXs 12 b

ROXs 12 is another neglected member of the Ophiuchus complex, located ∼1° south of its core. Bouvier & Appenzeller (1992) determined a spectral type of M0 and identified it as a young star based on the presence of weak Hα emission (EW[Hα] = −1.2 Å), X-ray emission, and lithium absorption. The HR diagram position of ROXs 12 corresponds to a mass of 0.87 ± 0.08 M_☉ and an age of 7.6$^{+4.1}_{-2.5}$ Myr. A significant excess in the WISE W3 and W4 filters indicates that the system hosts an optically thick protoplanetary disk, although the nondetection of emission at 1.3mm by (Andrews & Williams 2007) (F_ν < 19 mJy) suggests that the disk is not very massive (M_disk < 4 M_Jup). Also, the Hα line strength is consistent with most WTTSs and much lower than is seen for typical accreting stars. Ratzka et al. (2005) found no evidence of any close (≲1'') binary companions from speckle interferometry, a result confirmed and extended to much smaller radii by our observations with non-redundant aperture-mask interferometry (A. Cheetham et al., in preparation). Ratzka et al. (2005) did note a faint candidate companion at a projected separation of ∼17 (210 AU) in the shift-and-added stack of their speckle interferometry data. However, it subsequently has been neglected in the literature.

We obtained additional observations of ROXs 12 in 2011 and 2012, confirming the existence of the candidate companion (Figure 4, left). The multi-epoch astrometry demonstrates that the candidate companion has a proper motion similar to that of ROXs 12, with a relative motion of 7 ± 3 mas yr⁻¹ (4.2 ± 1.6 km s⁻¹) (Figure 4, right). The companion has dereddened J − H and H − K' colors very similar to those of ROXs 42B b, but is significantly redder at longer wavelengths, with K' − L' = 1.46 ± 0.11. While the measurement of comovement is more discrepant than for the other two new PMCs, our results rule out non-movement at high significance. If the candidate was unassociated, it therefore would be a nearby field star and the red colors of this candidate clearly distinguish it from the field population. Based on its absolute magnitude of $M_{K^{\prime }} = 8.42 \pm 0.19$, we estimate a companion mass of 16 ± 4 M_Jup for system ages of 1–5 Myr (Chabrier et al. 2000). As a bound companion of approximately planetary mass, we denote the companion as "ROXs 12 b."

Zoom In Zoom Out Reset image size

As for FW Tau b, the presence of a disk in the system (determined from the WISE W3 and W4 photometry) could indicate that we have observed a binary companion that is obscured by an edge-on disk. Given the lack of accretion onto the apparent primary of the system, the argument must be given special consideration for ROXs 12 b. However, as we discussed in Section 4.1.1, the point source morphology seen in our NIR images presents the same argument against this interpretation as for FW Tau b. The system is also quite luminous in W3 (m_W3 = 5.99), indicating that the companion would need to be massive and intrinsically luminous. The disk therefore is most likely associated with the primary star and the presence or absence of a disk around the companion cannot be inferred from the existing data.

As we show in Figure 5, six of the candidate companions that we have identified have relative motions that are similar to those expected for a non-moving background star. We therefore denote these apparent background sources as DoAr 22 cc1, GSC 06793-00994 cc1, GSC 06794-00156 cc1, PDS 70 cc1, UScoJ1608-1935 cc1, and ScoPMS 42b cc1 and have referred to them as such in this work. In most cases, the agreement with comovement is not exact, indicating that the candidate companion has a non-zero absolute proper motion, but these proper motions are generally ≲10 mas yr⁻¹.

Zoom In Zoom Out Reset image size

In Figure 6, we show the relative motions of four more candidates that appear to be moving with absolute motions of similar magnitude as (but in a different direction than) their host stars, indicating that they are likely not as distant as the candidates listed above. One target (hereafter denoted GSC 06191-00019 cc1) follows a uniform vector that is perpendicular (and of similar magnitude) to that of its candidate host star. Another target (ScoPMS 214) actually appears to be nearly comoving with its candidate host star, but past spectroscopic observations indicate that it is likely a field dwarf in the solar neighborhood (Metchev & Hillenbrand 2009). Finally, ScoPMS 42b hosts a pair of faint sources (separated by ∼150 mas) that appear to be comoving with each other, but not comoving with the candidate host. These sources were discovered (but not individually resolved) by Köhler et al. (2000), who found relative astrometry that is also consistent with the tracks that we observe. We infer that these objects (denoted ScoPMS 42b cc2 and ScoPMS 42b cc3) are a binary system comprised of two similar-mass field dwarfs.

Zoom In Zoom Out Reset image size

Finally, there are two additional targets for which we cannot determine a proper motion, but which we can classify as unassociated by other criteria. Both candidate companions (2M0415+2813 cc1 and HD 27659 cc1) were identified when we conducted snapshot AO imaging (in preparation for future deep observations) of newly identified Taurus members identified by Luhman et al. (2009) and Rebull et al. (2010). We concluded from the snapshot imaging that 2M0415+2813 cc1 was most likely a background galaxy, based on an elliptical elongation of its PSF that is not aligned with the P.A. to the host star (i.e., anisoplanatism) or with the elevation angle (i.e., windshake or dispersion). Subsequent to this determination, images from SDSS DR9 (Ahn et al. 2012) detected the candidate in the g' and r' filters with a contrast of 4–5 mag. Since late-M dwarfs have colors g' − K' ≳ 10 (Kraus & Hillenbrand 2007b), we conclude the candidate companion is indeed an unassociated background source. The other candidate companion, HD 27659 cc1, has low-quality astrometry from the discovery observations since the candidate fell on a high point in NIRC2's spatially correlated read-noise pattern; even though we have imaging observations from separate seasons, a test of common proper motion is inconclusive. However, the color of the companion is quite blue (J − L' = 0.12 ± 0.13), so it is unlikely to be a PMC. The proper motion of HD 27659 is also inconsistent with Taurus membership, so even though its position and distance place it within the Taurus molecular cloud (Kenyon et al. 1994), we suggest that the system is unlikely to be young.

We also have observed five known PMCs in order to obtain more or better colors than were available in the literature, as well as to extend the time baseline in confirming common proper motion. While their identity was not in doubt, we still have analyzed the literature data and our new observations in the same way as for the candidate PMCs (Sections 4.1–4.2). All are once again confirmed to be associated and we did not find any photometric measurements that contradict measurements in the literature.

GQ Lup continues to show significant evidence of orbital motion in the radial and tangential directions, as first pointed out by Neuhäuser et al. (2008). None of the other targets show clear evidence of motion at more than ∼3σ, which is not a sufficient criterion for significance among such a large sample of measurements. However, the circular orbital velocities at the observed separations (∼100–300 AU) are typically 1–2 km s⁻¹ or 1–2 mas yr⁻¹. We therefore expect that these measurements will achieve statistical significance in the near future, especially with dedicated campaigns that use a single instrument/filter and control or minimize systematic sources of error such as differential chromatic refraction and tilt jitter.